Diagnosis of Burn Injuries Using a Molecular Model Developed Based on 6 Key Immunity-Related Genes

Abstract

Background: Burn injuries are closely associated with significant morbidity and mortality. Furthermore, immune and distributive shock, often correlated with severe burns, can lead to multiple organ failure. Therefore, this study aimed to investigate differentially expressed genes (DEGs) and develop a diagnostic model for burn injuries. Method: RNA-seq data and clinical information of both burn injury and control samples in GSE19743, GSE77791, and GSE37069 datasets were obtained from Gene Expression Omnibus (GEO). DEGs were identified using the limma package. Moreover, functional enrichment analysis was performed utilizing the WebGestaltR package, while protein-protein interaction (PPI) analysis was conducted using a Cytoscape package. A novel diagnostic model was developed employing the least absolute shrinkage and selection operator (Lasso) regression and eXtreme Gradient Boosting (XGBOOST) package using the data from the GSE77791 dataset. Result: A total of 1950 DEGs, 1033 DEGs, and 865 DEGs were identified between burn injury and control groups in GSE19743, GSE77791, and GSE37069, respectively. The DEGs were closely associated to immune of burn patients. Gene set enrichment analysis (GSEA) revealed that four immune-related pathways were impeded in the burn injury group. Moreover, the burn injury patients persistently exhibited an immunosuppressed state. PPI analysis determined 13 hub genes, six of which were used to establish diagnostic model. Furthermore, the model showed high accuracy, sensitivity, and specificity with F1-Score (all 1). Additionally, the model showed an area under the receiver operating characteristic (ROC) curve (AUC) of 1 in the GSE77791 dataset, which was successfully validated in the GSE19743 dataset. Conclusions: This study determined 13 hub genes associated with immune, six of which were used to develop a diagnostic model for predicting the status of burn injury patients.